➀ Sensetime has secretly spun off its chip business after layoffs, attracting widespread attention.

➁ The company has completed billion-level financing to alleviate financial pressure.

➂ Sensetime has been seeking breakthroughs in the AI chip field since 2018.

➀ HP's servers are experiencing a surge in sales due to the adoption of NVIDIA's 'world's strongest AI chip', the GB200;

➁ Major cloud service providers like Microsoft and Meta are actively introducing GB200 and expanding their purchases of the higher-end NVL72 cabinets;

➂ Foxconn, as the exclusive supplier of GB200 and NVL72 cabinets, is seeing a flood of orders and a significant boost in server business.

➀ Hanwenji's stock price surged to a historical high, surpassing 140 billion yuan in market value, during the recent A-share bull market.

➁ Despite the company's consistent losses and declining revenue, the market's enthusiasm for Hanwenji is driven by its AI chip technology and the rise of AI and domestic substitution trends.

➂ Hanwenji, as a domestic AI chip leader, faces challenges such as funding and supply chain issues, but also has significant growth potential in the AI chip market.

➀ The increasing complexity of AI models and the exponential growth of network numbers and types have led to a dilemma for chip manufacturers between fixed function acceleration and programmable accelerators;

➁ The general AI processing methods are not up to standard, and focusing on specific use cases or workloads can achieve greater power saving and better performance in smaller space;

➂ The trend of AI algorithm complexity is increasing, and the number of floating-point operations is increasing, and the trend line is only pointing upwards and to the right.

➀ The Dimensity 9400, with a 3nm process, PC-grade Arm V9 architecture, and the 8th generation NPU, marks the latest achievement in mobile AI chips.

➁ The Dimensity 9400 supports the first-end DiT architecture, enabling video generation without internet connection on smartphones.

➂ The Dimensity 9400 achieves 32K text length for large models, eight times the length of the Dimensity 9300.

➃ The Dimensity 9400 supports running multi-modal large models on the endpoint and surpasses previous SOTA with a speed of 50 Tokens per second.

➄ The Dimensity 9400 leads the Zurich ETHZ mobile SoC AI performance list with a score of 6773, 1.4 times that of the Dimensity 9300.

➅ The Dimensity 9400 adopts TSMC's second-generation 3nm process, with a 35% single-core performance improvement and a 28% multi-core performance improvement over the previous generation, and a 40% reduction in power consumption.

➆ Mediatek has integrated the Dimensity AI Intelligent Agent Engine, which can achieve cross-application operations based on user needs and remember user habits, making the phone smarter and more convenient.

➀ The partnership between NVIDIA and TSMC, one of the most profitable in AI chip business, is showing signs of tension due to production issues with NVIDIA's new Blackwell chip.

➁ Issues with the chip, including potential design flaws, were discovered during testing by NVIDIA engineers.

➂ TSMC employees believe that NVIDIA's rushed production requirements contributed to the problems.

➃ NVIDIA's Blackwell chip mass production plan is delayed to the fourth quarter, with additional costs incurred.

➀ Computing power is an important indicator of a computer's information processing capability, with AI computing power focusing on AI applications, commonly measured in TOPS and TFLOPS, and provided by dedicated chips such as GPU, ASIC, and FPGA for algorithm model training and inference.

➁ AI chip accuracy is a way to measure computing power level, with FP16 and FP32 used in model training, and FP16 and INT8 used in model inference.

➂ AI chips typically use GPU and ASIC architectures. GPUs are the key components in AI computing due to their advantages in computation and parallel task processing.

➃ Tensor Core, an enhanced AI computing core compared to the parallel computation performance of Cuda Core, is more focused on the deep learning field and accelerates AI deep learning training and inference tasks through optimized matrix operations.

➄ TPUs, a type of ASIC designed for machine learning, stand out in high energy efficiency in machine learning tasks compared to CPUs and GPUs.

➀ YiZhu Technology announced the completion of several hundred million yuan in financing, led by a well-known overseas fund;

➁ The company is focused on innovative architecture to break and reduce 'storage wall', 'energy consumption wall', and 'compilation wall', developing products for data centers, AI cloud computing, and central AI servers;

➂ Since its establishment, YiZhu Technology has received extensive attention from the capital market, including support from major investment institutions.